A Sociolinguistic Profile of Kolami Dialects

DigitalResources Electronic Survey Report 2015-014

Compiled by Frank Blair A Sociolinguistic Profile of Kolami Dialects

Compiled by Frank Blair

Researchers: Frank Blair Bruce Cain Kathy Cain Jonathan Chauvan Devadanam Edwin Kiran Gaikwad Steve Watters

SIL International® 2015

Abstract

This survey was undertaken over a six-month period in late 1986 and early 1987 in five districts of India: four in eastern Maharashtra—Yavatmal, Wardha, Nanded, and Chandrapur District—and one, Adilabad District, in what was at the time western Andhra Pradesh.1 The survey had three main goals: 1) determine dialect areas, 2) assess the level of bilingualism, and 3) determine language use and attitudes. On the basis of wordlist comparisons, the survey established that there are at least two distinct dialects of Kolami: Eastern (or Southeastern) Kolami and Western (or Northwestern) Kolami. Dialect intelligibility testing confirmed these results. The wordlist comparisons further suggested the possibility of a third, Central, dialect. Using recorded text tests, the survey assessed the level of bilingualism among the Kolavar in the state languages of Marathi and Telegu. The results showed that most Kolavar are not sufficiently bilingual, with the exception of one community in Yavatmal District. As a result, literacy work would need to be conducted in the Kolam varieties. Finally, a questionnaire was administered to determine language use and language attitudes. It found that Kolami is being used widely and people have a positive attitude towards their language. In terms of language development, the survey recommends that separate projects are necessary for the Eastern and Western Kolami dialects. Speakers of Central Kolami may be able to use Western Kolami materials. [This survey report written some time ago deserves to be made available even at this late date. Conditions were such that it was not published when originally written. The reader is cautioned that more recent research may be available. Historical data is quite valuable as it provides a basis for a longitudinal analysis and helps us understand both the trajectory and pace of change as compared with more recent studies.—Editor]

1Editor’s note: Western Andhra Pradesh became a separate state in June 2014, known as Telangana. References to Andhra Pradesh in this report reflect the administrative divisions at the time of the survey in 1986–1987.

Contents

List of tables Preface 1 Introduction 1.1 Geography 1.2 People 1.2.1 Language 1.2.2 Population 2 Goals 2.1 Demographic profile 2.2 Linguistic similarity 2.3 Dialect intelligibility 2.4 Bilingualism 2.5 Language use and language attitudes 3 Summary of findings 3.1 Dialect areas 3.2 Bilingualism 3.3 Language use and language attitudes 4 Dialect areas 4.1 Linguistic similarity 4.1.1 Procedures 4.1.2 Sources of Kolami vocabulary 4.1.3 Similarity charts 4.2 Dialect intelligibility 4.2.1 Procedures 4.2.2 Dialect intelligibility scores 4.3 Conclusion 5 Bilingualism 5.1 Procedures 5.1.1 Pilot tests 5.1.2 Recorded Text Tests 5.1.3 Self-evaluation questionnaire 5.2 Preliminary bilingualism testing 5.3 Comprehensive bilingualism testing 5.4 Conclusion 5.4.1 Telugu 5.4.2 Marathi 6 Language use and language attitudes 6.1 Procedures 6.1.1 Questionnaires 6.1.2 Observation 6.2 Data 6.2.1 Sample 6.2.2 Language use 6.2.3 Language vitality 6.2.4 Language attitudes 7 Recommendations 7.1 For language development 7.2 For literacy 7.3 For further survey Appendix A: Demographic questionaire iii

Appendix B: Bilingualism questionnaire Appendix C: Language use and language attitude questionnaire References

List of tables

Table 1. Sources of Kolami vocabulary, in percentages (from Emeneau 1955:139) Table 2. Similarity percentages for three Kolami wordlists and a Naiki wordlist Table 3. Percentages of similar words in three Kolami wordlists and a Naiki wordlist Table 4. Similarity percentages for seven Kolami wordlists Table 5. Number of similar words out of total words in seven Kolami wordlists Table 6. Similarity percentages between Emeneau’s wordlist and three Western Kolami wordlists Table 7. Number of similar words between Emeneau’s wordlist and three Western Kolami wordlists Table 8. Similarity percentages between Emeneau and a Central and Eastern Kolami wordlist Table 9. Number of similar words between Emeneau and a Central and Eastern Kolami wordlist Table 10. Similarity percentages for four Kolami wordlists and Naik’s wordlist Table 11. Number of similar words in four Kolami wordlists and Naik’s wordlist Table 12. Similarity percentages for Rao’s wordlist and two Kolami wordlists Table 13. Number of similar words between Rao’s wordlist and two Kolami wordlists Table 14. Similarity percentages for Burrows and Bhattacharya’s wordlist and two Kolami wordlists Table 15. Number of similar words for Burrows and Bhattacharya’s wordlist and two Kolami wordlists Table 16. Four analytical situations on dialect intelligibly tests Table 17. Kolami dialect intelligibility test scores Table 18. Marathi recorded text test scores (pilot tests) Table 19. Test scores from Ner Taluk of Yavatmal District (broad categories) Table 20. Test scores from Ner Taluk of Yavatmal District (narrow categories) Table 21. Sample for Kolami language use and attitude questionnaire Table 22. Responses to question 8 by educated and uneducated people Table 23. Responses to question 9 by educated and uneducated people

Preface

This survey was undertaken in the last quarter of 1986 and the first quarter of 1987. It sought to discover more about the social and linguistic situation of the Kolavar in Yavatmal and the surrounding areas. It is with a feeling of gratitude that I acknowledge my indebtedness to the many persons who have contributed directly or indirectly to the present study. Without all this help and collaboration the result would not be what it is. It goes without saying that I alone am responsible for any errors of fact or opinion. The spelling of Marathi proper nouns is somewhat problematic when using Roman script. The ambiguity of the choices involved is increased by the fact that the proper nouns themselves often have more than one form, especially since the “Marathization” program has begun. I have tried to use the currently accepted Roman spellings of Marathi proper nouns insofar as I am aware of them. Where the current spelling differs more than marginally from previously common spellings, I have given the alternate spelling in parentheses the first time the word is used. Thereafter I have used the current spelling, except when quoting an author who uses one of the older spellings.

F. Blair, M.A. (Linguistics) Kathmandu, Nepal 15 November 1987

1 Introduction

1.1 Geography

Five districts were visited in the course of this survey. Four of these are in eastern Maharashtra: Yavatmal District (Yeotmal), Wardha District, Nanded (Nander) District, and Chandrapur District. The fifth district, Adilabad District, is in western Andhra Pradesh. Adilabad is bounded on the north by Yavatmal District and Chandrapur District, and on the west by Nanded District. Wardha District lies north of Yavatmal District.2 Topographically, this portion of the Andhra Pradesh-Maharashtra border alternates rolling hills with broad plains. Some of the hill areas in Chandrapur and Nanded are still forested, but deforestation has taken a toll in other regions.

1.2 People

1.2.1 Language

Among the many ethnic groups inhabiting this portion of the Andhra Pradesh-Maharashtra border are the Kolavar (or Kolams). The Kolavar speak a Dravidian language known as Kolami. Emeneau (1955:141) says: “That Kolami is a language of the Dravidian family hardly needs detailed demonstration.” On the basis of his field inquiries, Emeneau (1955) lists four Kolami dialects. One is spoken in Wardha District and in northern Yavatmal District (in Yavatmal taluk). A second is spoken by the Kolavar of Kelapur and Wun taluks, also in the Yavatmal District. (This presumably includes the areas now called Maregaon and Wani taluks.) A third dialect is spoken by the Kolavar of Andhra Pradesh. The Kolavar of Chinnur and Sirpur taluks of Adilabad District are said to have a dialect different from that of the other Kolavar in Andhra Pradesh, being more influenced by Telugu, but it is not clear if the Kolavar of these two taluks have given up the Kolami language entirely in favor of Telugu. A fourth dialect is said to be spoken by the Kolavar of Betul District of Madhya Pradesh, though Emeneau says he has no other report (besides his field inquiries) of the existence of Kolavar in Betul District (ibid.:4). Emeneau (idem) also reports that he was told that the Kolavar of Wardha District and Yavatmal taluk in Yavatmal District are said to form one endogamous division, and the Kolavar of Kelapur and Wun taluks form a second, the Kolavar of Andhra a third, and those of Betul a fourth. This survey did not establish the existence of Kolavar in Betul District. The district was visited several times in the course of the research for Stahl 1986 and no Kolavar were encountered. None of the Kolavar interviewed in the course of the research for this project knew of any Kolavar north of Pulgaon taluk in Wardha District. There is a population of Korku people in Betul District speaking a Munda language; perhaps these were confused with the Kolavar at some point. The researchers were not able to visit Chinnur taluk in Adilabad District to confirm or deny the existence of a separate Kolami dialect in that region either. This still remains to be done. There do seem to be three endogamous divisions among the Kolavar, coinciding with those described by Emeneau apart from Betul. The Kolavar of western Yavatmal District and Wardha District generally do not intermarry with those of Adilabad District. Nor do the Kolavar of eastern Yavatmal intermarry with those of Adilabad. The reason usually given for this is that they belong to different social groups. The Kolavar of western Yavatmal generally do not intermarry with those of eastern

2Editor’s note: The original file contained three maps which could not be reproduced here. The hardcopy file of the survey report including the maps is available from SIL Archives.

Yavatmal, but the reason given for this is that it is too far away, which makes it difficult to arrange marriages. There seems to be no objection offered on the grounds that the Kolavar of Yavatmal belong to two different social groups. A distinction is drawn between Kolavar living in the hills and those living on the plains. This is most clearly seen in Kinwat taluk of Nanded District. In Kinwat, the plains Kolavar refer to themselves as “Madka” Kolavar. They have marriage relations with the Kolavar of western Yavatmal. The hill Kolavar refer to themselves as “Metla” Kolavar in this region. Their marriage relationships are with the Kolavar of Adilabad District. According to Von Fürer-Haimendorf (1945:4), the Kolavar of Kinwat are largely detribalized and “Marathi-speaking.”3 This is an odd statement in light of the fact that Emeneau and others found vital Kolavar communities in Kinwat. Emeneau (1955:140) even goes so far as to say that the Kolavar of West Adilabad (in Andhra Pradesh!) may have had more access to Marathi than the Kolavar of Kinwat. It appears that Von Fürer-Haimendorf must be referring to the Madka Kolavar, while Emeneau refers to the Metla Kolavar, for the Madka are more bilingual in Marathi than the Metla. Emeneau does, however, say that “in general, what is on record for bilingualism suggests that practically all adult Kolams speak a subsidiary language” (ibid:5). This “subsidiary language” varies from place to place and may be either Marathi, Telugu, or Gondi, though not all Kolavar can speak Gondi. Emeneau says that the variety of Marathi used by the Kolams of Wardha District is a “rustic” variety, which is not easily intelligible to speakers of standard Marathi, though it is commonly used by mother tongue speakers of Marathi in the area. Emeneau goes on to say that his (standard) Marathi-speaking interpreters told him that “the Kolams are not very much at home in Marathi of any kind when there is any departure from the most frequent subjects of everyday discourse” (ibid:5). It is clear from this that the Kolavar of Maharashtra are not very bilingual in Marathi. Concerning bilingualism among the Kolavar of Andhra Pradesh, Rao (1950:6) says that Telugu speakers refer to the Kolavar as Mannyods, a term which is used by Kolams in east Adilabad (i.e., Sirpur and Chinnur taluks). These, he says, no longer speak Kolami, but only Telugu. He also says, Almost all the Kolams in Adilabad, excepting the taluks of Sirpur and Chinnur, know Gondi. On the other hand very few Gond know Kolami....In Sirpur...the Kolams do not know Gondi....The Gonds and the Kolams converse with each other in Telugu in Sirpur taluk....In Sirpur and Chinnur...the Kolams either talk Telugu or Teluguised Kolami. (Rao 1950:5ff)4 It is hard to know how much weight to place on these somewhat contradictory generalizations. The same applies to the following claim by Rao: Excepting the Kolams of Sirpur and Chinnur those living in the other Talukas of Adilabad speak a fairly uniform Kolami language....More than three-fourths of the Kolams speak the language (of the western Kolams in Adilabad)….Since the majority of the Kolams are to be found in the adjoining District of Yeotmal [Yavatmal], and speak the dialect used by the West Adilabad Kolams, this dialect can be taken to be the representative Kolami. (Rao 1950:7) Rao’s identification of the Kolami of Adilabad with the Kolami of Yavatmal is not accurate. Emeneau (1955:2) is also skeptical of the claim that the Kolami dialect of Adilabad is the same as that of Yavatmal. He notes that data from eastern Yavatmal is more similar to his Wardha data, which differs from that of Rao, than it is to Rao’s data.

3I have not seen Von Fürer-Haimendorf 1945. The material quoted here is gleaned from Emeneau 1955. 4I have not seen Rao 1950. The material quoted here is gleaned from Emeneau 1955.

1.2.2 Population

According to the Project Report on Primitive Tribes (hereafter PR) by the government of Maharashtra (n.d.:17), the tribal population of Yavatmal District is 197,977. Of these, 49,977 are Kolams. Yavatmal taluk has 157 villages with Kolam habitations for a total Kolam population of 21,219; Kelapur taluk has 123 villages with Kolam habitations for a total Kolam population of 15,845; Wani taluk has 107 villages with Kolam habitations for a total Kolam population of 12,711 (PR n.d.:16) The report is not dated, but internal evidence makes it seem likely that it was published either in the late seventies or the early eighties. It appears that the figures quoted from it reflect the 1971 Census. Information about the number of Kolam in Adilabad is more difficult to come by. According to Von Fürer-Haimendorf: Neither Kolams nor Naikpods have been recorded adequately in the Census of 1941 when they were erroneously classified as subtribes of the Gonds. Only 746 Kolams were returned, but I estimate that there must be at least 3,000 or 4,000 Kolams in Adilabad District, and probably more if the Marathi-speaking and largely detribalized Kolams of Kinwat are counted. (1945:page unknown) Rao (1950:1) says that the number of Kolavar in Andhra Pradesh is “at least, if not more than five thousand.” His figures refer primarily to Adilabad District. According to him Sirpur taluk alone has two thousand Kolams. The number of Kolavar living in Chandrapur, Wardha, and Nanded Districts is not known, but it is clear that most Kolavar live in Yavatmal District. Perhaps Chandrapur and Adilabad contain about the same number of Kolavar, with Nanded ranking fourth, and Wardha fifth in terms of Kolavar population. The Kolavar of Nanded appear to be concentrated exclusively in Kinwat taluk. It seems likely that a conservative estimate of the number of Kolavar would exceed 60,000 people. The report gives the following information concerning the Kolavar of Yavatmal District. The literacy rate for Yavatmal District as a whole is 31 percent. For tribals it is 13 percent, and for Kolavar it is 4 percent (PR n.d.:17). According to the report (n.d.:34), “Primary education is a farce in Kolam villages. Kolam children are shown on the rolls of the schools but they hardly attend.” Concerning other aspects of Kolavar life, the report says: Details regarding the occupational distribution of Kolams are not available. However, it is estimated from the records that 26 percent Kolams (sic) are engaged in agriculture; 27 percent as agricultural laborers; 7 percent as forest laborers and 40 percent as non-workers including dependents....Poor transport and communication facilities are a great deterrent factor for the development of the Kolam villages....Average Kolam family earns Rs. 500 to 600 a year. (PR n.d.:17–22) It may be seen from this that the Kolavar of Yavatmal District are among the more isolated and educationally disadvantaged ethnic groups in the district. The situation does not differ radically for the Kolavar in other districts.

2 Goals

2.1 Demographic profile

First, the survey will investigate the distribution of Kolavar communities using Kolami as the traditional mother tongue in the Yavatmal District of Maharashtra as well as the surrounding districts.

2.2 Linguistic similarity

Second, the survey will explore the degree of linguistic similarity among the varieties of Kolami spoken by the Kolavar and to compare Kolami with other Dravidian languages spoken in the area. The degree of linguistic similarity will be expressed as a percentage based on a comparison of equivalent words using standard wordlists.

2.3 Dialect intelligibility

Third, the survey will assess dialect intelligibility among any local languages that might appear to be similar to Kolami. The degree of inherent intelligibility will be expressed as a percentage based on the average of the scores on a simple language test.

2.4 Bilingualism

Fourth, the survey will investigate bilingualism among the Kolavar. It will assess what segments of the Kolavar population are able to understand oral narrative texts in Marathi and Telugu. This will be done using self-evaluation questionnaires and simple language tests. Population segments will be described in terms of the major social criteria which affect bilingualism. For the purposes of this report, bilingual ability will be evaluated in terms of the following continuum.

<------+------+------+---> Nonfunctional Functional Competent

If a community or individual is said to be competent in a language, it means that their ability to communicate in that language is nearly equivalent to that of a mother-tongue speaker. A competent speaker of a language can—and regularly does—use that language in the same way and the same domains that a mother-tongue speaker does. An individual or community described as having functional ability in a language may be sufficiently skilled in the language to use it regularly in certain limited domains, but has neither the ability nor the desire to use the language for all purposes. For example, people with functional ability in a language may be able to understand oral narrative texts in that language but prefer to use another language for everyday communication. If an individual or community is described as nonfunctional in a language, it means that their ability to use that language is marginal or non-existent. Such people may know enough of the language to accomplish some simple tasks in the language, but it is not a language they regularly use.

2.5 Language use and language attitudes

Finally, the survey will investigate the language use situation in the Kolavar communities. Particular attention will be paid to determining the language used in the home, traditional activities, and village pursuits. Attitudes towards the various languages spoken in the region will also be investigated. Language use and language attitudes will be investigated using informal questionnaires and observations. The attitudes expressed will be described in terms of the following continuum.

<---+------+------+------+------+------+------+----> Strongly Positive Mildly Neutral Mildly Negative Strongly Positive Positive Negative Negative

3 Summary of findings

This chapter provides a summary of the findings for dialect areas, bilingualism, and language use and language attitudes. These three topics will be discussed in more detail in the chapters that follow, including methodology and presentation of the data.

3.1 Dialect areas

Not all the varieties of Kolami are inherently intelligible with each other. There are phonological differences between the varieties used in the different parts of the area surveyed, and these seem to affect the ability of people to understand each other. It is quite clear that there are two different dialects of Kolami. One of these is spoken in Adilabad District, Chandrapur District, and Nanded District. In this report it is referred to as Eastern or Southeastern Kolami. The number of Eastern Kolami speakers is in the order of ten thousand. The second dialect is called Western Kolami or Northwestern Kolami in this report as it is spoken in Wardha District, Yavatmal District, and Nanded District. The number of speakers of Western Kolami is about fifty thousand. Eastern Kolami and Western Kolami are inherently unintelligible. A third dialect of Kolami is spoken in and around Wani and Maregaon taluks of Yavatmal District. It is called Central Kolami in this paper. It is inherently intelligible with Western Kolami. Its speakers have been enumerated with those of Western Kolami, as it is unclear exactly where the boundary between these two dialects lies. Some authors, notably Rao (1950), point to the possibility of a fourth Kolami dialect spoken in Chinnoor and Sirpur taluks of Adilabad District. There is no independent confirmation of this. There are other areas of Adilabad District from which information is needed before it can be stated with certainty that Eastern Kolami is the only Kolami dialect spoken there.

3.2 Bilingualism

The majority of Kolavar in Maharashtra are nonfunctional in either Telugu or Marathi, the state language of Andhra Pradesh and Maharashtra respectively. Information concerning the Kolavar in Adilabad District is not complete enough to determine the bilingual situation of the Kolavar of Andhra Pradesh.

3.3 Language use and language attitudes

Attitudes towards their mother tongue vary somewhat from individual to individual and from community to community among the Kolavar. Attitudes toward the various regional languages vary similarly. Generally speaking, Kolavar attitudes toward their mother tongue range from mildly negative to mildly positive, depending on the domain in focus. The mother tongue is the language of choice for intracaste communication. In those domains which require intercaste communication, the state language, either Marathi or Telugu, is the language of choice for those able to communicate in it. The state languages are being increasingly recognized as the languages necessary for pursuing non-traditional occupations. The most common domains for use of the mother tongue by the Kolavar is at home with family and with other Kolavar, or at work, if the work is agricultural and the co-workers are also Kolavar. The most common domain for the use of the state language is at the bazaar. Attitudes toward the state languages range from neutral to very positive. In most cases the state language is the language of wider communication. This has encouraged some individuals to try to use the state language in every sphere of language use. This phenomenon does not appear to be common, and in most cases attitudes towards the state language are more accurately described as neutral. Most Kolavar think of Kolami as the appropriate language for the home, though some individuals encourage their children to acquire the state language as a second language.

4 Dialect areas

4.1 Linguistic similarity

4.1.1 Procedures

4.1.1.1 Wordlist elicitation

The degree of linguistic similarity among the various Kolavar communities was determined by eliciting wordlists in each community. These wordlists were then compared with each other in order to determine the percentage of similarity among them.5 The standard procedure for eliciting a wordlist involved collecting a 210-item wordlist from a speaker in one community and then checking it with another speaker from the same community. In Kolavar communities it was necessary to check wordlists carefully in order to distinguish between Telugu or Marathi words which have become part of the Kolami vocabulary and those which have not but were offered because Telugu or Marathi was the language used in elicitation. Some lists are more “contaminated” than others. Checking the wordlists was also necessary in order to ensure that the correct words were elicited. It was not always possible to follow this procedure strictly. In some communities it was not possible to obtain the entire wordlist; in others it was not possible to check the wordlist with more than one speaker. Factors which prevented our thoroughly checking or eliciting entire wordlists included constraints on time as well as apathy or even hostility on the part of the community. Both of these factors somewhat lessened the reliability of the wordlists in assessing the degree of linguistic similarity in some cases. For this reason it is not necessary for the percentage figure expressing degree of similarity to be 100 percent for the two wordlists to be considered as representing the same dialect. The threshold figure used in this report is 80 percent. In other words, if 80 percent of the words on a wordlist are considered phonologically similar to those on another wordlist, then the two dialects are considered to be the same. Setting the limit at 80 percent seemed to confirm what mother-tongue speakers said about their perceptions of the similarity or difference between the Kolami dialects within their experience. In some cases the 80 percent threshold also reinforced conclusions drawn from intelligibility testing, which seems to confirm its validity for use in cases where there are no data from intelligibility data. For a discussion of the standard problems involved in eliciting and comparing wordlists, see Grimes 1988.

4.1.1.2 Wordlist comparison

Linguistic similarity was evaluated by comparing the 210-item wordlists elicited in each community with those elicited in other communities. The results of this comparison are expressed as a percentage of phonologically similar words. It should be noted that the percentage of similarity arrived at in this manner is not the same as a percentage of cognates. Cognate percentages may be discovered only by the application of the comparative method. This is beyond the scope of this project and so a comparative analysis of these Kolami dialects has not been attempted. Therefore, the percentages in this report are not percentages of cognate words. The criteria used for determining the similarity of two wordlists are subjective ones based on presumed ease of communication. C. Rensch and D. Marshall (1987, personal communication) suggest using the phonological similarity of the corresponding words of a wordlist as an index to ease of

5The wordlists appear in the original, printed version of this report. This is available from SIL Archives but is in an older font.

7 communication. The next three paragraphs are a prose summary of the criteria that have been used in determining whether a pair of words is phonologically similar. Two words with the same meaning are considered to be phonologically similar if a certain percentage of the sounds (phonetic segments) occurring in those words are the same or similar. Two sounds are considered to be the same if 1) they may be represented by the same phonetic symbol, 2) in the case of vowels, if they differ by only one phonetic feature, or 3) if they are phonetically similar and correspond in at least three pairs of the words being compared. On the other hand, two sounds are considered to be similar if they are phonetically similar but do not correspond in three or more pairs of words. For the purpose of the comparison, vowels which differ from each other by two or more phonetic features may also be counted as similar if they correspond in at least three pairs of words, even though such vowels might not normally be considered to be phonetically similar. Furthermore, in order for two words to be considered similar, at least half of the sounds in the longer of the two words must correspond to the same sounds in the shorter of the two words. At least another quarter of the sounds must correspond to similar sounds in the shorter of the two words. The remaining sounds may correspond to sounds which are not phonetically similar or to no sound at all. If the words being compared consist of only two sounds, then both sounds must be the same. If they consist of three sounds, then two of the sounds must be the same and the third must be similar. The percentage of similarity between two wordlists is an expression of the number of phonologically similar words in the two wordlists. Each word in a wordlist is compared with the corresponding word in the second wordlist. The number of phonologically similar pairs of words is then divided by the total number of comparisons. The resulting number, multiplied by one hundred, is the percentage of similarity.

4.1.2 Sources of Kolami vocabulary

Table 1 presents the sources of Kolami vocabulary as tentatively analyzed by Emeneau (1955). Western Kolami is the wordlist he elicited in Wardha District. It consists of 931 words. The Eastern Kolami wordlist was elicited for the present survey in Adilabad District. It consists of 652 words. The Kinwat wordlist was elicited in the Nanded District. It is more similar to the Adilabad wordlist than it is to the Wardha wordlist and thus may be said to also represent Eastern Kolami. It consists of 435 words.

Table 1. Sources of Kolami vocabulary, in percentages (from Emeneau 1955:139)

Dravidian Indo-Aryan Unknown Western Kolami (Wardha) 55 35 10 Eastern Kolami (Adilabad) 62 22 16 Kinwat (Nanded) 78 11 11

Emeneau (1955) believes that most of the Indo-Aryan vocabulary which has been borrowed is drawn form Marathi. Concerning the greater percentage of Dravidian material items in the Kinwat wordlist, he says: Certainly, materials collected in a few hours of field work are always likely to fail in randomness for the reason that the investigator may prefer to get material that may be useful for historical studies, rather than a really random sample which might not prove to be quite so useful historically. In other words, he does not think that the fact that his analysis of Kolami vocabulary shows different percentages of source material is a clear guide to dialect differences. He says that the reason for this is that the sample is biased by the fact that he was eliciting material for a historical analysis.

Emeneau also discusses the relationship of Kolami with three other Dravidian languages: Parji, Naiki, and Ollari. Parji may be the Dhurwa dialect spoken in Bastar District of Madhya Pradesh. Naiki may refer to a dialect of a Kolavar subcaste in Adilabad District. The subcaste itself is apparently also known as Naiki in some parts of Andhra Pradesh.6 Emeneau says that these three Dravidian languages— Parji, Naiki, Ollari—are “closely related” to each other as well as to Kolami. He adds: The group shows closer sub-relationships of Kolami and Naiki on the one hand and of Parji and Ollari on the other. The relationship of Kolami and Naiki is very close. It seems possible on the basis of such information as is at hand that they are easily mutually intelligible (i.e., that they are dialects of one language); the very few minor differences that appear...can hardly make against intelligibility. Specific information on this matter would be welcome. The relationship of Parji and Ollari is less close, as is that between either of them and Kolami-Naiki, and it seems doubtful whether mutual intelligibility can mark any of these relationships. (Emeneau 1955:142) It is clear from this passage that Emeneau does not expect Ollari and Kolami, or Parji and Kolami, to be mutually intelligible. If Naiki is the dialect of a segment of the Kolavar community of Adilabad District, then it may very well be mutually intelligible with Eastern Kolami, which is also spoken in Adilabad. Emeneau does give a short Naiki vocabulary. Table 2 is based on the comparison of the words in this Naiki wordlist with Kolami wordlists from three different taluks, one from Western Kolami and two from Eastern Kolami. The Pulgaon wordlist represents Western Kolami while the Utnur and Asifabad wordlists represent Eastern Kolami. Table 3 shows the numbers of comparisons that were made to arrive at each percentage in table 2. Table 2. Similarity percentages for three Kolami wordlists and a Naiki wordlist

Pulgaon (Western) 79 Naiki Utnur 68 88 (Eastern) Asifabad 61 85 86 (Eastern) Table 3. Number of similar words in three Kolami wordlists and a Naiki wordlist

Pulgaon (Western) 97/123 Naiki Utnur 89/131 70/80 (Eastern) Asifabad 119/194 101/119 105/122 (Eastern)

On the basis of these wordlist comparisons in tables 2 and 3, it is clear that Naiki should be grouped with Eastern Kolami rather than Western Kolami, confirming our conclusion above that Naiki may be mutually intelligible with Eastern Kolami.

6The term “naik” is generally used to refer to a leader, and, sometimes, to the members of a dominant subcaste within a caste. Thus many castes and tribes, including the Kolavar in some regions, have a group to which this term is applied. According to D. Marshall (1987, personal communication) there are Naikis in Adilabad District who speak a Dravidian language distinct from Gondi. Perhaps these are the people represented by the wordlist which Emeneau refers to as Naiki. If so, then it seems likely that they represent a subcaste of the Kolavar.

4.1.3 Similarity charts 4.1.3.1 Eight newly elicited Kolami wordlists Kolami wordlists were elicited from Kolavar speakers living in the Adilabad District of Andhra Pradesh, and in the Wardha, Yavatmal, and Nanded Districts of Maharashtra. The majority of the Kolavar live in these four districts. In the Adilabad District, the wordlists were elicited from Kolavar living in Utnur taluk and Asifabad taluk; in the Wardha District from Kolavar living in the Pulgaon taluk; in Yavatmal District from Kolavar living in the Ner, Maregaon, and Wani taluks; and in the Nanded District from Kolavar living in Kinwat taluk. Two wordlists were elicited in the Kinwat taluk of Nanded District because there are two distinct communities of Kolavar here: the Metla Kolavar live in the forested and hilly areas of the taluk, while the Madka Kolavar live on the plains. In total, eight wordlists were elicited in the course of the survey: two from Adilabad District, one from Wardha District, three from Yavatmal District, and two from Nanded District. Table 4 presents the results of the comparison of seven of these wordlists. Table 5 shows the number of comparisons that were made to arrive at each percentage in table 4. The wordlist from Ner taluk is 94 percent (186/198) similar to the Pulgaon and Kinwat (Madka) wordlists. Thus, as it appears to represent the same variety of Kolami as the Pulgaon and Kinwat (Madka), it has not been included in table 4. Table 4. Similarity percentages for seven Kolami wordlists Kinwat (Metla) (Nanded) Utnur 83 (Adilabad) Asifabad 80 86 (Adilabad) Maregaon 69 76 70 (Yavatmal) Wani 68 74 67 89 (Yavatmal) Kinwat(Madka) 69 71 67 83 83 (Nanded) Pulgaon 65 68 61 79 78 93 (Wardha)

Table 5. Number of similar words out of total words in seven Kolami wordlists

Kinwat (Metla) (Nanded) Utnur 109/131 (Adilabad) Asifabad 156/194 105/122 (Adilabad) Maregaon 142/207 100/131 137/186 (Yavatmal) Wani 136/201 96/130 128/190 180/202 (Yavatmal) Kinwat (Madka) 142/206 91/129 128/192 171/205 165/199 (Nanded) Pulgaon 135/208 89/131 119/194 163/207 157/201 192/206 (Wardha)

From the information presented in table 4 and table 5, it is clear that there are at least two dialects of Kolami. One dialect, Eastern Kolami, is spoken in the southern and eastern regions of the Kolami area (and thus also known as Southeastern Kolami). It is represented by the wordlists from the Metla Kolavar of the Kinwat taluk in Nanded District, and the Utnur and Asifabad taluks of Adilabad District. The other dialect, Western Kolami, is spoken in the northern and western regions of the Kolami area (and thus also referred to as Northwestern Kolami). It is represented by the wordlist from the Madka Kolavar of the Kinwat taluk in Nanded District and the Pulgaon taluk in Wardha District, as well as the Ner taluk in Yavatmal District (which, as we noted above, is 94 percent similar to the other two but not included in tables 4 and 5). The percentage of similarity between Western and Eastern Kolami wordlists is generally less than 70 percent. The Kolami spoken by the Metla and Madka Kolavar of Kinwat taluk typifies the dissimilarity between Eastern and Western Kolami. The degree of similarity between these two varieties of Kolami is only 69 percent. Though these two communities live in the same taluk, their speech is quite different from each other. The wordlist elicited from the Metla Kolavar shows much more similarity to the Kolami spoken in Adilabad District (80 and 83 percent) than it does to the Kolami spoken in Yavatmal District (68 and 69 percent). Conversely, the wordlist elicited from the Madka Kolavar shows much more similarity to the Kolami spoken in Yavatmal District (83 percent) than it does to the Kolami spoken in Adilabad District (69 and 71 percent). On the basis of this evidence it is clear that there are at least two Kolami dialects: Eastern (or Southeastern) and Western (or Northwestern). The information in table 4 also indicates the possible presence of a third Kolami dialect, which we may refer to as Central Kolami. The wordlists elicited in Maregaon taluk and in Wani taluk of Yavatmal District may represent this third dialect. The Kolami of these two taluks is definitely different from the Eastern Kolami dialect as the percentage of similarity is always below 80 percent. On the other hand, the wordlists from Wani and Maregaon bear a closer relationship to the Western Kolami dialect as represented by the wordlists elicited in Pulgaon and Ner taluks and from the Madka Kolavar of Kinwat taluk, the percentage of similarity always being more than 75 percent. The degree of similarity between Pulgaon and Kinwat is 93 percent, while the highest percentage of similarity of the Kinwat list with those of Maregaon and Wani is ten percentage points lower, at 83 percent. The Pulgaon list is even less similar to the Maregaon and Wani lists, 78 and 79 percent. It is clear that there is a degree of dissimilarity between the varieties of Kolami spoken in these two areas (Western and Central Kolami), but it is difficult to say how important this dissimilarity is based on information from wordlists alone.

4.1.3.2 Four previously published Kolami wordlists

There are several previously published Kolami wordlists. Of these, perhaps the most useful and reliable is the one which appears in Emeneau 1955. In addition, Emeneau presents not only the words elicited in the course of his fieldwork in Wardha District, but also Kolami wordlists from other (published and unpublished) sources, especially Rao 1950. Other sources of Kolami words are the wordlists by Burrow and Bhattacharya (1960) and Naik (1973). These four wordlists were compared with those elicited by the researchers in order to discover what degree of similarity they represented.

1. Emeneau’s wordlist

The fieldwork presented in Emeneau 1955 was carried out in the Wardha District of Maharashtra. Table 6 presents the percentage of similarity of the words he elicited to our own wordlists from the Western Kolami dialect. Table 7 shows the number of comparisons that were made to arrive at each percentage in table 6. As may be seen from table 6, the wordlist extracted from Emeneau 1955 is similar to the Western Kolami dialect. In the case of Pulgaon this is not surprising in light of the fact that Pulgaon taluk lies in the same district where Emeneau elicited his wordlist, although the figures for the Madka in Kinwat taluk of Nanded District and the Ner taluk in Yavatmal District are even higher.

Table 6. Similarity percentages between Emeneau’s wordlist and three Western Kolami wordlists

Pulgaon (Wardha) Emeneau 85 (Wardha) Kinwat (Madka) 93 88 (Nanded) Ner 95 88 94 (Yavatmal) Table 7. Number of similar words between Emeneau’s wordlist and three Western Kolami wordlists Pulgaon (Wardha) Emeneau 139/163 (Wardha) Kinwat(Madka) 192/206 141/161 (Nanded) Ner 189/199 140/159 186/198 (Yavatmal)

On the other hand, as table 8 shows, Emeneau’s wordlist is only 73 percent similar to Eastern Kolami (represented by the wordlist taken among the Metla Kolavar of Kinwat taluk in Nanded District) and 77 percent similar to Central Kolami (represented by the wordlist from the Wani taluk in Yavatmal District). Table 9 shows the numbers of comparisons that were made to arrive at each percentage in table 8. Table 8. Similarity percentages between Emeneau and a Central and Eastern Kolami wordlist

Wani (Yavatmal) Emeneau 77 (Wardha) Kinwat (Metla) 68 73 (Nanded) Table 9. Number of similar words between Emeneau and a Central and Eastern Kolami wordlist

Wani (Yavatmal) Emeneau 123/160 (Wardha) Kinwat (Metla) 136/201 119/163 (Nanded)

2. Naik’s wordlist

The fieldwork presented in Naik 1973 was carried out in the Yavatmal District of Maharashtra. Table 12 compares the percentage of similarity of the words he elicited to the wordlists elicited for this survey from the Central Kolami dialect (Wani, Maregaon, and Ner in Yavatmal District) and the Western Kolami dialect (Pulgaon in Wardha District). Table 11 shows the numbers of comparisons that were made to arrive at each percentage in table 10.

Table 10. Similarity percentages for four Kolami wordlists and Naik’s wordlist

Wani (Yavatmal) Naik 83 (Yavatmal) Maregaon 89 83 (Yavatmal) Pulgaon 78 82 79 (Wardha) Ner 85 84 86 94 (Yavatmal)

Table 11. Number of similar words in four Kolami wordlists and Naik’s wordlist

Wani (Yavatmal) Naik 142/171 (Yavatmal) Maregaon 180/202 146/176 (Yavatmal) Pulgaon 157/201 143/175 163/207 (Wardha) Ner 164/194 153/183 171/200 187/199 (Yavatmal)

Naik’s wordlist is similar to those elicited by the researchers in the Wani, Maregaon, Pulgaon, and Ner taluks. The others range between 79 and 94 percent similar to each other; Naik’s similarity percentages range between 82 and 89 percent. This means that Naik’s wordlist is representative of the Western and Central Kolami dialects. Apparently the items in Naik’s wordlist were elicited from people in more than one taluk. This may account for the fact that this wordlist shows about the same degree of similarity with Western Kolami (82 percent) as with Central Kolami (83 and 84 percent), assuming that the latter is indeed distinct from Western Kolami. While this is not shown in tables 10 and 11, Naik’s wordlist is only 69 percent similar (121/175) to the wordlist elicited among the Metla Kolavar of Kinwat taluk in Nanded District. Naik’s wordlist appears to be quite distinct, therefore, from Eastern Kolami.

3. Rao’s wordlist

The fieldwork presented in Rao 1950 was carried out in the Adilabad District of Andhra Pradesh. Table 12 compares the percentage of similarity of the words he elicited to two wordlists from this survey representing the Eastern Kolami dialect: from Utnur taluk in Adilabad District and from the Metla Kolavar of Kinwat taluk in Nanded District. Table 13 shows the numbers of comparisons that were made to arrive at each percentage in table 12.

Table 12. Similarity percentages for Rao’s wordlist and two Kolami wordlists

Utnur

(Adilabad) Rao 84 (Adilabad) Kinwat (Metla) 83 88 (Nanded)

Table 13. Number of similar words between Rao’s wordlist and two Kolami wordlists Utnur (Adilabad) Rao 77/92 (Adilabad) Kinwat (Metla) 109/131 126/143 (Nanded)

As table 12 shows, Rao’s wordlist is very similar to the Eastern Kolami dialects represented by the wordlists elicited in Utnur taluk and among the Metla Kolavar of Kinwat taluk. Though not shown here, it is only 69 percent similar (98/143) to the Western Kolami dialect represented by the Pulgaon wordlist. The degree of similarity with the Central Kolami is slightly higher at 72 percent (100/138).

4. Burrows and Bhattacharya’s wordlist

The fieldwork presented in Burrows and Bhattacharya 1960 was carried out in the Kinwat taluk of Nanded District of Maharashtra. Table 14 presents the percentage of similarity of the words they elicited to wordlists representing the Eastern and Western Kolami dialects spoken by the Metla and Madka Kolavar communities in Kinwat taluk respectively. Table 15 shows the number of comparisons that were made to arrive at each percentage in table 14.

Table 14. Similarity percentages for Burrows and Bhattacharya’s wordlist and two Kolami wordlists

Kinwat (Metla) (Nanded) Burrows and 88 Bhattacharya (Nanded) Kinwat (Madka) 69 64 (Nanded)

Table 15. Number of similar words for Burrows and Bhattacharya’s wordlist and two Kolami wordlists Kinwat (Metla) (Nanded) Burrows and 75/84 Bhattacharya (Nanded) Kinwat (Madka) 142/206 54/84 (Nanded)

Table 14 shows that the Kolami dialect studied by Burrows and Bhattacharya in Kinwat taluk is much more similar to the Eastern Kolami dialect of the Metla Kolavar (88 percent) than it is to the Western Kolami dialect of the Madka Kolavar (64 percent).

4.2 Dialect intelligibility

4.2.1 Procedures

Intelligibility among Kolami dialects was determined using the procedures set forth in Dialect Intelligibility Testing by Eugene Casad (1974). In most ways we followed his procedures closely. The most significant

14 departure concerns the method of formulating questions for the texts. Casad recommends that a group of mother-tongue speakers examine the text and create questions for it. Then the group chooses the best questions of those generated for use in the test. Instead, the researcher formulated the questions after completing both a literal and an idiomatic translation of the text selected for testing. These questions were then translated back into the local dialects. Because this procedure increases the likelihood of producing unusable questions while creating intelligibility tests, we always created substantially more than ten questions. Questions which were consistently misunderstood by mother-tongue speakers of the dialect were eliminated from the tests, and they did not count in figuring the percentage of intelligibility. The physical location where the researchers gathered stories and turned them into test tapes are called reference points. From these reference points the stories are then played for people at the various test points to see how well they are understood. Every test tape is first checked by testing it at the place where it was collected to make sure the people there can understand it. This is called the hometown or control test. A typical sample size for dialect intelligibility testing is ten people. When analyzing the results of dialect intelligibility tests it is important to understand the role of standard deviation in interpreting the average percentage score listed. If the standard deviation is small (ten or less), that means that the spread of the scores (the difference between the highest and the lowest score) is relatively small. In other words, all the test subjects generally scored close to the average score. If the standard deviation is high (above fourteen or so), that means that the spread of the scores was much greater, showing that some people understood the story on the test tape much better than others and so scored higher. Four analytical situations are reflected by the various combinations of average score and standard deviations. These are shown in table 16. If the average is high (above 80 percent) and the standard deviation is low (under ten), then we can say the story on the test tape is fairly easily understood in that area. There is not much difference among the individual scores. If the average is low (below about 70 percent) and the standard deviation is also low, then it means that almost no one understood the story on the test tape. Any time the standard deviation is high, no matter whether the average score is high or low, then we know that people’s understanding of the story on the test tape varied widely from person to person. This is often the case when some people in a community have learned to speak another language, but others have not. This kind of understanding is different from that more uniform understanding which results from two languages being inherently similar.

Table 16. Four analytical situations on dialect intelligibly tests

Standard deviation High Low

Situation 1 Situation 2 Many people understand the Most people understand the

High story on the test tape well, but story on the test tape. some have difficulty. Situation 3 Situation 4

Many people cannot understand Few people are able to Average score Average Low the story, but a few are able to understand the story on the answer correctly. test tape.

4.2.2 Dialect intelligibility scores

Table 17 summarizes the results of the dialect intelligibility tests. Reference points (along the top) are the places where the test tapes were made. Test points (along the side) are the places where the tapes were tested. Ner and Pulgaon represent Western Kolami. Wani represents Central Kolami. Kinwat (Metla) and Rajura represent Eastern Kolami. For each test, the percentage that a community scored (x) is followed by the standard deviation (s) and the sample sizes for the community (n).

Table 17. Kolami dialect intelligibility test scores

Reference Points Ner Wani Kinwat Rajura Pulgaon (Metla) Ner x= 95 91 60 58 s= 7.1 12 9.4 10 n= 10a 10 10 10 Wani x= 92 62 63 91 s= 9.4 10 11 9.9 n= 12 10 10 10

Kinwat (Metla) x= 50 98 95 39 s= 21 3.9 5.3 13

Test points n= 8 17 8 9 Rajura x= 55 97 99 54 s= 10 4.7 3 13 n= 11 11 11 11 Pulgaon x= 97 s= 4.8 n= 10 x = average score; s = standard deviation; n = sample size aThe original has “12” here, which appears to be a typo.

The results of the dialect intelligibility testing between Kinwat(Metla) and of Rajura are characteristic of situation 2 in table 16: low deviation (4.7 and 5.3) and high average score (97 and 95). It is clear that they are part of the same Eastern Kolami dialect area. The results of the dialect intelligibility testing between Wani and Western Kolami (represented by Ner and Pulgaon) in table 17 are ambiguous. The average scores are high (91), and the standard deviations are also fairly high (9.9 and 12), which is characteristic of situation 1 in table 16. It is possible that speakers of Central and Western Kolami understand each other fairly well because of contact between the two communities and not because the two dialects are inherently intelligible. The Eastern Kolami dialect of Rajura and Kinwat (Metla) is not inherently intelligible with the Western Kolami dialect of Ner or Pulgaon and the Central Kolami dialect of Wani. The result of dialect intelligibility testing among these dialects is characteristic of situation 3 in table 16: high standard deviations (ranging from 9.4 to 13) and low average scores (ranging from 39 to 62).

4.3 Conclusion

The wordlist comparisons in table 4 show that the wordlists of Kinwat (Metla), Utnur, and Asifabad are more similar to each other than any of them are to Maregaon, Wani, Pulgaon, or Kinwat (Madka). And the reverse is true as well: the wordlists of Maregaon, Wani, Pulgaon, and Kinwat (Madka) are all more similar to each other than any of them is to the wordlists from Utnur, Asifabad and Kinwat (Metla). The dialect intelligibility tests confirmed this. It is clear, therefore, that the Eastern Kolami dialect spoken in

Adilabad district and by the Metla Kolavar of Kinwat is not inherently intelligible with the Western and Central Kolami dialect of Wani, Ner, Pulgaon, Wardha, and of the Madka Kolavar of Kinwat. Even though the results of the dialect intelligibility tests were ambiguous in this regard, on the basis of the wordlist comparisons in table 4, the Kolami dialect spoken in Maregaon and Wani is slightly different from the dialect of Pulgaon to the northwest and of the Kinwat (Madka) to the southeast. The Kolami dialect of Maregaon and Wani may, therefore, be thought of as a third dialect, Central Kolami.

5 Bilingualism

5.1 Procedures

5.1.1 Pilot tests

The first step in this bilingualism study is pilot testing of various Kolavar communities on narrative texts in the relevant state languages in order to get a rough index of the level of bilingualism. Recorded text tests were used to determine the level and extent of bilingualism in Telugu and Marathi, the state language of Andhra Pradesh and Maharashtra, respectively. The procedures involved in recorded text testing are described in Casad 1974 (see also section 4.2 above on dialect intelligibility testing). Results of recorded text tests are expressed as a percentage of questions on the tests which are answered correctly by the respondent. The results of the pilot testing determined that many Kolavar were not functional in their ability to use either Telugu or Marathi. However, a Western Kolami-speaking community in Ner taluk of Yavatmal District appeared to be more than usually proficient in Marathi, and to have more contact with Marathi speakers because of its proximity to Yavatmal, the district headquarters.

5.1.2 Recorded Text Tests

The pilot test of the Kolavar community mentioned in the previous section, who live in and around the villages of Lamina and Chichbardi in Ner taluk of Yavatmal District, resulted in a score higher than 70 percent in Marathi. For this reason it was deemed necessary to conduct a more extensive bilingualism investigation in this community. This evaluation formed the second step of the bilingualism study. Two recorded text tests were administered to the sample. The first was in the local Kolami dialect and ensured that the sample was competent in the test taking procedure. If an individual did not score above 70 percent on the test in the local dialect, that person was excused from the sample and a replacement was found. The second test was the one used in the aforementioned pilot testing. This text was recorded by a mother-tongue speaker of Marathi. The variety of Marathi used is the educated standard. This Marathi text was control tested with a group of ten mother-tongue speakers of Marathi.

5.1.3 Self-evaluation questionnaire

In addition to recorded text testing, a self-evaluation questionnaire was administered to the same sample of individuals as those who took the recorded text tests. This self-evaluation questionnaire requires people to evaluate their ability to perform certain tasks involving differing degrees of complexity using Marathi. The use of self-evaluation questionnaires and the procedures for their construction is described in Grimes 1986, Quakenbush 1986, and Blair 1987. A copy of the questionnaire used to evaluate people’s bilingual ability in the relevant state language is contained in the appendix to this report. Unfortunately, the questionnaire turned out to be an inappropriate means of evaluating bilingual ability among the Kolavar. There are two reasons for the failure of the questionnaire. In the first place, most people responded positively to most of the questions on the bilingualism self-evaluation questionnaire. Since most people indicated that they could perform all the tasks in Marathi described in the questionnaire, the results were not useful in discriminating levels of bilingual ability.

The second reason for the failure of the questionnaire also has to do with the undiscriminating nature of the responses given by the people interviewed in the course of the study. It is very difficult to formulate a questionnaire that investigates something other than what people want the researcher to think they are able to do with the second language. This makes questionnaires useful in investigating such language posture, but will result in their being invalid in certain investigations of bilingual ability. For example, in the case of the Kolavar of Ner taluk, it is important for them to be bilingual in Marathi, as they have a great deal of contact with Marathi speakers. Some of the people being interviewed were thus motivated to overstate their bilingual ability. In one case, an older woman who had responded positively to all the questions on the questionnaire scored only 55 percent on the Marathi recorded text test while she had scored 90 percent on the test in her own language. Results of self-evaluation questionnaires may be expressed in one of two ways. Sometimes the results are expressed in terms of the percentage of questions that a respondent answers positively with regard to bilingualism. In other cases the results are expressed in terms of Second Language Oral Proficiency Evaluation (SLOPE) levels. SLOPE rates bilingual proficiency on a six-point scale, from Level Zero (no proficiency in the second language) to Level Five (proficiency equal to an educated mother- tongue speaker of the language). As the results of the self-evaluation questionnaire among the Kolavar are inconclusive, they are not discussed in this report.

5.2 Preliminary bilingualism testing

The Marathi text test scores are presented in table 18.7

Table 18. Marathi recorded text test scores (pilot tests)

Test points Average Standard Sample deviation Ner 79% 8 9 Wani 51% 23 12 Kinwat (Metla) 67% 26 10 Rajura 31% 14 11 Pulgaon 69% 11 10

The results of the pilot testing determined that many Kolavar were not functional in their ability to use either Marathi or Telugu, with the exception of the Western Kolami-speaking community in Ner taluk of Yavatmal District mentioned above, which was very proficient in Marathi.

5.3 Comprehensive bilingualism testing

Table 19 contains the results of the bilingualism evaluation done in Ner taluk of Yavatmal District, which scored highest on the Marathi pilot tests (see table 18). Recorded test texts in Kolami (K) and Marathi (M) were administered to the Kolavar communities living in and around the villages of Lasina and Chichbardi. The Kolavar living in this area are among the most bilingual that researchers encountered in the course of the survey. These villages are only a few kilometers from the district headquarters and just off the main road. Many of the Kolavar living here work as day laborers in the district headquarters or for Marathi-speaking employees in the surrounding area.

7Editor’s note: The complete text test scores appear in an appendix to the original printed version of this report. This data is available from SIL Archives but is in an older font which could not be reproduced here.

Table 19. Test scores from Ner taluk of Yavatmal District (broad categories)

Average Sample Standard (x) Size (n) Deviation (s) Sample K 95 51 6.4 M 85 51 12 Men K 95 33 6.7 M 84 33 11 Women K 96 18 6.2 M 85 18 15 Literate K 94 15 7.4 M 89 15 12 Illiterate K 96 36 6.1 M 83 36 12 Young (<30) K 94 22 8 M 86 22 12 Middle (31–45) K 96 18 5 M 86 18 9.9 Older (>45) K 95 11 5.2 M 80 11 26

Table 19 presents the test results according to three broad social categories—age, sex, and ability to read—as reported by the people being interviewed. Fifty-six people were interviewed in the course of the bilingual evaluation. One person was not willing to take the recorded text tests. Another person was neither from Ner taluk nor living there. Because he was only visiting, his scores were not counted in this sample. Three other people scored 70 percent or below on the Kolami test. Scores from these interviews were also eliminated from the sample, resulting in a sample size of fifty-one. It will be seen from table 19 that scores on the Marathi recorded text test average ten points less than the scores on the Kolami home town test. A recorded text test on a narrative text is an indicator only of very basic language ability. When used as an indicator of bilingual ability, even scores as high as 85 percent must be interpreted to mean that the group in question is not adequately bilingual. It is doubtful whether recorded text tests are able to give any indication of bilingual ability higher than Level Two on the SLOPE scale. A score of 100 percent on a recorded text test might indicate that a group was Level Two (or higher) on the SLOPE scale. Scores less than 90 percent should be interpreted to indicate that the group being evaluated is less than Level Two on the SLOPE scale. From table 19 it appears that the illiterate portion of the sample is less bilingual than the literate portion of the population. The ability to read and write usually corresponds with the number of years in school. In this sample most of the Kolavar who reported themselves to be illiterate also said that they had spent less than four years in school. That portion of the sample which reported itself to be literate appears to be the most bilingual part of the sample. According to the scores in table 19, older people form the least bilingual segment of the sample. This is partly due to the fact that no literate older people were found. The older portion of the sample therefore consists entirely of illiterate people. In most of the categories in table 19, the standard deviation of the scores on the Marathi test is fairly high, usually 12 percent or higher. This indicates that in each portion of the sample, a part of that group scored significantly better than average and another part scored significantly worse. Even within the broad categories delineated by table 19, bilingual ability appears to vary somewhat. In an attempt to isolate the reasons for this variation, the scores for the sample presented above were analyzed according to narrower social categories than those presented in table 19. The results of this analysis may be found in table 20.

Table 20 presents the scores on the recorded text test in terms of twelve different social categories. Each of these categories is represented by a box in the chart in table 20. An attempt was made to administer recorded text tests to at least five people in each of these categories but this was not always possible. For example, no older, literate Kolavar of either sex were tested. The researchers did not encounter any older literate people in the course of the bilingualism evaluation. It was difficult to persuade women to be tested, so they are underrepresented in several categories. In categories where the number of responses obtained is too few to be statistically significant, the raw scores obtained from the tests that were administered are listed in parentheses.

Table 20. Test scores from Ner Taluk of Yavatmal District (narrow categories)

Young Middle Older (under 30) (30–45) (over 45) Kolami Marathi Kolami Marathi Kolami Marathi x= 95 81 97 86 93 82 Illiterate s= 8.4 11 4.6 8.7 5.2 13 n= 6 6 10 10 6 6 Men x= 93 84 (90) (100) Literate s= 8.9 13 (100) (82) n= 8 8 (90) (100) x= 94 89 95 82 96 77 Illiterate s= 8.9 12 5.8 13 5.5 21 n= 5 5 4 4 5 5 Women x= (90) (91) (100) (82) Literate s= (100) (100) n= (100) (100)

The information presented in table 20 shows that older illiterate women and young illiterate men are among the least bilingual portions of the sample. The younger illiterate men scored on average eight points lower than women of similar age and educational attainments. Middle-aged men, however, are marginally more bilingual than women of the same age.

5.4 Conclusion

5.4.1 Telugu

None of the Kolavar communities studied in the course of the bilingualism evaluation is significantly bilingual in Telugu. It was not possible to carry out evaluations in Telugu everywhere that might be deemed useful, particularly in Andhra Pradesh. However, it is clear that Kolavar communities in Maharashtra are not appreciably bilingual in Telugu.

5.4.2 Marathi

Most of the Kolavar communities studied in the course of the bilingualism evaluation are not very bilingual in Marathi. Most Kolavar in rural areas typically control only enough Marathi to visit the bazaar and meet basic needs in that language. Kolavar living in towns and cities, those with some education, and those who work as day laborers for Marathi-speaking employers have a higher degree of bilingual ability in Marathi. However, even among the Kolavar in these categories, there are many who are not above Level Two on the SLOPE scale.

6 Language use and language attitudes

6.1 Procedures

The primary technique used to gather information on language use and language attitudes was a questionnaire (see appendix to this report). This questionnaire was administered while gathering information for the demographic profile or while administering recorded text tests. A second technique, observation, was used when possible.

6.1.1 Questionnaires

Informal questionnaires are a useful way of discovering information about the attitudes of people, about the way they think things should be. They are particularly useful in uncovering information about language posture, that is, about what people want you to think they do, and what you wish them to think about various languages. A person who does not know Marathi may say that Marathi is the best language to use when trying to get a job. The fact that he does not know Marathi, and that he himself uses his Kolami dialect on the job is irrelevant in this context; his statement shows a very good attitude toward Marathi in some economic situations.

6.1.2 Observation

The second technique used to gather information about language use and language attitudes is simply observation. By observing people speaking in a variety of different contexts one can usually determine which languages are considered appropriate in different situations. By listening to people as they talk to each other, an observer can often determine which languages are considered appropriate to talk about certain subjects. Observation is particularly useful in finding out what people actually do, assuming the observer has the opportunity to observe language use in a wide variety of situations. As such it provides a way of verifying information gathered with questionnaires. If an observer sees several people applying for work and notes that Marathi is the language used when one speaks to an employer, then he has gained useful information about the situations in which Marathi is used.

6.2 Data

6.2.1 Sample

Observation and pilot testing for bilingualism showed the communities living in and around the villages of Lasina and Chichbardi in Ner taluk of Yavatmal District to be more bilingual in Marathi than many other Kolavar communities visited in the course of the survey. If language attitudes and use patterns show that Kolami is still a vital language in these two bilingual communities, then a similar conclusion may be drawn about less bilingual communities. Table 21 describes the sample which responded to the questionnaire with regard to age, education, and sex.8 The sample consisted entirely of mother-tongue speakers of Kolami who claimed Lasina and Chichbardi as their place of birth. Everyone interviewed said that they knew Marathi and used it either frequently or on a daily basis. Of the people interviewed, thirteen people said they were literate in Marathi.9 (For the questionnaire used to elicit this information, see the appendix to this report.)

8Editor’s note: There is an error in this table. The total sample size is said to be 45 but the numbers actually add up to 47. 9Editor’s note: Complete information on the sample may be found in an appendix to the original printed version of this report which is available from SIL Archives.

Table 21. Sample for Kolami language use and attitude questionnaire

Younger Middle Older (under 31) (31–45) (over 45) Educated 9 0 0 (standard 3–8) Men Uneducated 5 4 7 (no school) Educated 3 3 0 (standard 3–8) Women Uneducated 5 4 7 (no School)

6.2.2 Language use

The first five questions on the questionnaire (see appendix) are directed towards eliciting language use patterns. Questions one and two asked about language use in the home. Thirty-three people gave information about what language they spoke to their children. Of these, thirty-two said that they used Kolami. Twenty-six people responded to the question about what language their spouses used with children. All of them said that their spouses used Kolami. The number of total respondents to these two questions is less than 45 because not everyone in the sample was married or had children. These responses indicate that Kolami is by far the most commonly used language in the home. The third and fifth questions on the questionnaire are directed towards eliciting language use patterns when speaking with people who are not members of the Kolavar community. Everyone in the sample reported using Marathi to speak with people who were not Kolavar. Similarly, everyone in the sample reported that Marathi was the language they used at the bazaar. These responses indicate that Marathi is the language which is used to communicate with people who are not part of the Kolavar community. The fourth question asked whether there were ever occasions when they used Marathi to talk with other members of the Kolavar community. Out of the total sample of forty-five, only nine people reported occasional use of Marathi with other Kolavar. Thirty-six people said that they did not use Marathi to talk to other Kolavar. These responses indicate that Kolami is the language of preference for communication within the Kolavar community.

6.2.3 Language vitality

The sixth question on the questionnaire addresses language vitality. It asked the respondents if they knew of any Kolavar who could not speak Kolami. The question actually asked if they knew where there might be Kolavar who could not speak Kolami to rule out hypothetical answers. Only one person out of forty-five said that he knew of Kolavar that did not speak Kolami. Along with the responses to questions one and two, this indicates that the ability to speak in Kolami is not being lost. Instead it is being retained for use in the home and for use with other Kolavar while Marathi is used for other purposes such as market transactions.

6.2.4 Language attitudes

The last three questions on the questionnaire are directed towards eliciting attitudes toward the Kolami and Marathi languages. Question seven asked people to predict whether or not their children would speak Marathi better than Kolami. Of the twenty-five people who responded to this question, twenty-one said that their children would be able to speak Kolami better than Marathi. Similarly, in response to the ninth question, forty out of forty-five people said that Kolami is the best language for the Kolavar people.

Taken together, the responses to these two questions indicate a positive attitude towards Kolami on the part of the Kolavar people. The responses to the eighth question are somewhat problematic. This question asked whether Kolami or Marathi was easiest to learn. It was thought that a person who said that Kolami is easier to learn would probably hold a more positive attitude toward Kolami than a person who answered Marathi. However, only nineteen people out of forty-five said that Kolami was the easier of the two languages to learn. This means that some of the people who felt that Kolami was the best language for the Kolavar in question nine also felt that their language was more difficult to learn. There are (at least) two possible explanations for this. It may be that the answers to this question are based on the fact that in this community many Kolavar are able to speak some Marathi, while they rarely encounter mother-tongue Marathi speakers who have acquired Kolami as a second language. From this fact people may decide that since second-language speakers of Marathi are more common than second-language speakers of Kolami that Kolami is a more difficult language to acquire. Another explanation for the response to this question may be that Marathi is taught in school and there are text books and other materials available which make it possible to study Marathi while this is not true of Kolami. Although the eighth question did not ask which language was easiest to study in school it may have been understood that way by the people who answered the question. If this is the case then it seems likely that those people with some experience in school would say that Marathi is the easier of the two languages to “learn.” Table 22 gives the responses to the eighth question according to whether or not the respondents have been to school.

Table 22. Responses to question 8 by educated and uneducated people

Kolami Marathi Total Uneducated 17 13 30 Educated 2 13 15 Total 19 26 45 X2=6.07; df=1; p=.014

Table 22 shows that among the Kolavar who have never been to school, about half (seventeen) thought that Kolami was easier to learn than Marathi and about half (thirteen) thought that Marathi was easier. Among the Kolavar who had been to school, only two thought that Kolami was easier; the other thirteen thought that Marathi was easier to learn. This is a significant difference between educated and uneducated people in answering this question. According to a standard statistical test of significance known as the chi square test (X2 test), there is only a 1.4 percent chance that the difference in answering this question is accidental. This makes it seem likely that education is influencing the answer to this question and that it is not that people have negative attitudes toward Kolami. Table 23 matches the responses to question 9 (Which language is best for the Kolavar people?) with their educational background. It indicates that most Kolavar have good attitudes toward Kolami regardless of educational background.

Table 23. Responses to question 9 by educated and uneducated people

Kolami Marathi Total Uneducated 27 3 30 Educated 13 2 15 Total 40 5 45 X2=.03; df=1; p=.867

The results of the chi square test in the case of table 23 indicate that there is no correlation between level of education and the answer given to the ninth question. Both educated and uneducated Kolavar have equally favorable attitudes towards Kolami.

7 Recommendations

7.1 For language development

This survey has shown that the Eastern Kolami and Western Kolami dialects are not inherently mutually intelligible. Moreover, most of the Kolavar communities are not appreciably bilingual in either Marathi or Telugu. In the light of these two facts, the Kolavar communities of Maharashtra and Telugu need language development projects in their mother tongue. Separate projects for the Western and Eastern Kolami dialects are necessary. Speakers of Central Kolami and Western Kolami may be able to use the same materials. If speakers accept the materials already prepared in these dialects, separate projects should not be necessary. If acceptance is a problem, then further study of the relationship between Central and Western Kolami will have to be done.

7.2 For literacy

The same two facts cited in the previous section indicate that literacy programs in Kolami are appropriate. It seems likely that Kolavar learning to read in Marathi or Telugu would quickly find themselves unable to cope with the demands placed on them by the second language medium. This would lead to discouragement and high drop out rate in a Marathi or Telugu literacy program. Such an experience may serve to convince many people that because they cannot learn to read Telugu or Marathi, they cannot learn to read at all. A literacy program in Kolami avoids these problems. It is true that Kolami is not perceived as a language for use in the educational domain in the way that Marathi or Telugu is. This means that vernacular literacy promotion will play an important part in the literacy program. The advantages of a Kolami literacy program are self-reinforcing. The more people understand what they are learning to read, the more desirous they are of learning more. For this reason, literacy projects in Kolami are recommended.

7.3 For further survey

The primary need for further survey of Kolami dialects lies in Adilabad District. In particular, more knowledge of the bilingualism situation in that district would be useful, as well as a more thorough check for other Kolami dialects.

Appendix A: Demographic questionnaire

1. Number of interview 2. Date of interview 3. Location of interview 4. Name 5. Age 6. Sex 7. Education 8. Occupation 9. Previous occupation(s) 10. Religion 11. Place of current residence 12. Place(s) of previous residence 13. Caste or social class 14. Clan or moiety 15. Marital status 16. Number of children 17. Number of people in household 18. Mother tongue 19. Other language(s) 20. Literate in which scripts?

Appendix B: Bilingualism questionnaire

1. Original: Can you buy food in the market at a just price? Contextualized: Can you buy needed things at the bazaar? Kolami: niv angadita wasti vikat semeng shekilativ genang?

2. Original: Can you understand and respond correctly to questions about where you are from, your marital status, occupation, date and place of birth? Contextualized: Can you answer questions about your home, village, family, occupation? Kolami: niv inne allanet, umet, kulumbnet, dhandhanet, parasnanet utter shiyang shekelativ genang?

3. Original: Can you buy a needed item of clothing or a bus or train ticket? Contextualized: Can you tell the conductor where you want to get down and arrange for a bus ticket? Kolami: S.T. yeting digenglaga ade konduktorung edtive genang? tikit wellative genang?

4. Original: Can you give a brief story of your life and tell of immediate plans and hopes? Contextualized: Can you give a brief story of your life and talk about future plans? Kolami: immet jamlopal ghadilul gotti itdatir genang?

5. Original: Can you describe what types of leaders you have and what each one does in leading the people? Or can you describe the way children are taught what they need to know to become adults? Contextualized: Can you talk about what was said at the last Kolam panchayat meeting? Kolami: panchlopal erud gotti itdatir genang?

6. Original: Can you hire someone to work for you and arrange details such as salary, qualifications, hours, specific duties? Contextualized: If you wanted a job, can you discuss pay and duties with your employer? Kolami: imung dhanda lagtin te malik wetta gotti rajdatir genang?

7. Original: Can you defend your beliefs or those of your people against criticism from someone else? Contextualized: Can you scold/argue with someone who speaks Marathi? Kolami: kumbapana mudekar wetta niv kejakadatira?

8. Original: Are you afraid that you will misunderstand information given to you? Contextualized: If your child is very sick, can you explain to the doctor what is wrong? Kolami: immet bala wesang andan te dagdarakun weltoldatira?

9. Original: Do you find it difficult to follow and contribute to a conversation among native speakers who try to include you in their talk? Contextualized: If you listen to Marathi people speaking, can you understand everything they say? Kolami: imun wetta marathipana mudekaret pana kella genang?

10. Original: Can you use as many words in the language as in your language? Contextualized: Do you know as many Marathi words as Kolami words? Kolami: kolawa shebdul ondar aning marathi shebdul anar genang?

11. Original: Can you change the way you talk, depending on whether you are talking to educated people, close friends, those who work for you? Contextualized: Can you speak respectfully to an official and informally to a child? Kolami: niv sahebul wetta manad mudatir genag?

12. Original: Can you serve as an informal interpreter for a leader from your mother tongue group who may not be able to speak the second language very well? Contextualized: Can you translate for a friend who cannot speak Marathi? Kolami: yev marathi muda toter awur wetta marathi muddatira?

Appendix C: Language use and language attitude questionnaire

1 English: What language do you speak to your children? Kolami: niv inne balakul wetta taned bhashat mudtiv?

2. English: What language does your spouse speak to your children? Kolami: inne mahasal inne balakul wetta taned bhashat mudativ?

3. English: What language do you speak to non-Kolami people? Kolami: niv biger kolawarshiva taned bhash mudativa?

4. English: Do you ever speak Marathi to Kolami people? Kolami: niv kolawar wetta marathi mudatire genang?

5. English: What languages do you speak at the bazaar? Kolami: niv angadit taned bhasha mudativa?

6. English: Are there Kolami people who cannot speak Kolami? Where? Kolami: kolawa bhash wahtote intolok indargengeng, yetin?

7. English: Will your children be able to speak Kolami better than Marathi? Kolami: inne balakul kolawarung peksha owal marathi mudar genang?

8. English: Which language is easier to learn, Kolami or Marathi? Kolami: karpengsatti ed bhasha sope anda, kolami kinwa marathi?

9. English: Which language is better for Kolami people, Kolami or Marathi? Kolami: kolawarungsatti ed bhasha owal ande, kolami kinwa marathi?

References

Blair, F. 1987. Sociolinguistics on a shoestring: A survey manual. Unpublished manuscript. Burrow, T., and S. Bhattacharya. 1960. A comparative vocabulary of the Gondi dialect. Journal of Asian Studies 2:73–251. Casad, Eugene. 1974. Dialect intelligibility testing. Summer Institute of Linguistics Publications in Linguistics and Related Fields 38. Norman, OK: The Summer Institute of Linguistics of the University of Oklahoma. Emeneau, Murray B. 1955. Kolami: A Dravidian language. Annamalai University Annamalainagar book 8611 KTM IC Rs. 11.50. Berkeley, CA: University of California, Berkley. Government of Maharashtra; Department of Social Welfare, Cultural Affairs and Sports. n.d. Project report on primitive tribes: Kolams in the Yavatmal District. Grimes, Barbara F. 1986. Evaluating bilingual proficiency in language groups for cross-cultural communication. Notes on Linguistics 33:5–27. Grimes, Barbara F. 1988. Why test intelligibility? Notes on Linguistics 41:39–64. Naik, Titus L. 1973. Kolami word list. Patterns in clause, sentence, and discourse in selected languages of India and Nepal, vol. 4. Ronald Trail. Norman, OK: The Summer Institute of Linguistics of the University of Oklahoma. Quakenbush, John Stephen. 1986. Language use and proficiency in a multilingual setting: A sociolinguistic survey of Agutaynen speakers in Palawan, Philippines. Georgetown University. Ph.D. disseration. Rao, P. Setumadhava. 1950. A grammar of the Kolami language. Hyderabad. Stahl, James L. 1986. A sociolinguistic survey of the Korku language area. Unpublished manuscript. Von Fürer-Haimendorf, Christoph. 1945. Tribal populations of Hyderabad yesterday and today. Hyderabad. Reprinted from Census of India, vol. 21, pp. xix–xx. 1941. Government Central Press.